Recent HDR Works

[Hoang]

Let's have a recent HDR works thread  

Here's some stuff I shot and edited yesterday, using a Canon 20D and a Tokina 12-24 f/4
The HDR/tonemapping was done with Picturenaut, lens correction via PTLens, more toning via Alien Effects Exposure 2, then some dodging/burning + cloning in Photoshop

[Alaska]

Are you using the new Beta version of Picturenaut which now imports Canon RAW files?  Or the older version?

How many images were processed and at what EV?  Seems that the general recommendations are -2, 0 and +2.  But find that depending on the image an additional +4 and maybe -4 may be needed.

Giving both Dynamic Photo HDR and Photomatix Pro a try for light HDR post processing and then into PhotoShop for more work.

[Hoang]

Are you using the new Beta version of Picturenaut which now imports Canon RAW files?  Or the older version?

How many images were processed and at what EV?  Seems that the general recommendations are -2, 0 and +2.  But find that depending on the image an additional +4 and maybe -4 may be needed.

Giving both Dynamic Photo HDR and Photomatix Pro a try for light HDR post processing and then into PhotoShop for more work.

Hi, Alaska. Yes, I am using the beta version; however, I use ACR/Photoshop to batch convert the 20D RAWs into lens distortion corrected, defringed 16bit Profoto TIFFs which I import into Picturenaut.
I never use the Bracketing feature of my camera, instead, I use the histogram to increase my shutter speed (lower exposure) until there are no clipping highlights all (using UniWB to be sure) then bracket up manually every 2 stops (set my camera to 1/2 stops so that it takes only 4 wheel rotations) until the darkest shadows are approximately midtones (again, via using histogram). This allows me to have all the highlight detail I need but also lets me have very clean shadows (since the signal to noise ratio is better for midtones than for shadow values in a single shot).
To me, this makes much more sense than choosing some arbitrary "middle 0ev" shot and not being sure if you are saving the would-be clipped highlights and shadows.

[Alaska]

Will try your suggestion for the Canon on exposure.  Wish that it would do 5 in place of 3.  Hopefully the dials will not give out.  But if they do, then there is always the 50D for replacement!  

Do you find with Picturenaut that you get better tone mapping if first converting to 16 bit TIFFs in place of directly using RAW in Picturenaut?  However, as you pointed out, using TIFFs will give editing control for your application.

[Hoang]

Will try your suggestion for the Canon on exposure.  Wish that it would do 5 in place of 3.  Hopefully the dials will not give out.  But if they do, then there is always the 50D for replacement!  

Do you find with Picturenaut that you get better tone mapping if first converting to 16 bit TIFFs in place of directly using RAW in Picturenaut?  However, as you pointed out, using TIFFs will give editing control for your application.

I am not sure. The reason I prefer using TIFF is because I can do PTLens lens correction, ACR defringing, and set all the white-balances to be the same (as you stated), so I've never tried using RAWs directly into Picturenaut.

[Guillermo Luijk]

I do all the blending with Zero Noise and manual tone mapping in Photoshop usually with 2 curves. Shooting always with exposure bracketing {-2,0,+2} (if my camera allowed {-3,0,+3} I would probably use it instead), where -2 is an exposed to the right shot preserving highlights.
I never touch the camera and set mirror lockup while shooting for optimum alignment and sharpness.

From 2 shots 4EV apart:






From 3 shots 2EV apart:






More than 3 shots is seldom necessary and most times unadequate because blending software tends to do a progressive blending of the images losing more sharpness the more images are used, and for noise and DR more images are not needed at all.

[Geoff Wittig]

I do all the blending with Zero Noise and manual tone mapping in Photoshop usually with 2 curves.

Pardon my stupidity, but how do you perform manual tone mapping in Photoshop? Forgive me if this has been discussed in a previous  thread.

[Jann Lipka]

Guillermo
I think your samples look  very good, thanks .
As usual you are a great technician ( and creator ) .

Would you share with us your manual blending approach ?

I like your idea about  2 image mix - common sense says
that middle tones should be well represented in such a +/- 2 EV mix .
The fewer images to fiddle around with the better I think :-)

Also I think camera  manufacturers should give us ability to bracket with more then 2 EV .
Nikons max 1 EV is probably enough with Slide film ....
My 1 DS gives +/- 3 EV ...
Sometimes it is too much apart :-)

[Guillermo Luijk]

Would you share with us your manual blending approach ?

I like your idea about  2 image mix - common sense says

Well the blending is not manual, I do it with Zero Noise which provides a virgin unprocessed optimum blending of the N RAW files provided into an output TIFF file. This TIFF contains all the HDR information but needs proper tone mapping to display it, and this is what I do manually with 2 curves (usually): bright curve with mask to give extra protection to the highlights, and global contrast curve.

Image on the left is straight output from Zero Noise, image on the right after two curves:



You can download the original TIFF with the curves from here.

The number of needed shots for a proper HDR capture is a very funny subject of discussion. I have seen guys performing as many as 19 (yes 19) shots, thinking that would make the result better. That is absolutely a waste of time, storaging space, processing power and in the end, a loss of quality. Unless all the shots are perfectly aligned (which is not likely if needed to touch the camera), there will be more sharpness loss the more images you captured.
On the other hand more images does not improve anything. IMO 3 is the most balanced number, and we can set them 3EV apart with current cameras.

A camera has a DR of about 8-f-stops of range acceptably free of noise. If our shots are separated 3EV, that means after an optimum blending we are only going to use about the 4 upper f-stops of the sensor's DR, which are the highest quality f-stops. We don't need more.
Shooting: -3,-2,-1,0,+1,+2,+3 will provide nearly the same amount of information as -3,0,+3, but will need more than double space, capture time and processing time.